[adwn]

I think you're missing the main argument: It would take a lot of money and cross-domain knowledge to design and build a passable printer from scratch. Contrast that with software, where all you need is time and a low to medium power computer. In fact, compared to software open source, all other open source engineering is non-existent (relative to OSS, not in absolute terms).

[criddell]

Why would you design one from scratch? The hardware out there is excellent and inexpensive. Start with that and write the firmware in a similar way that the Magic Lantern group did for cameras.

[bgorman]

Embedded systems generally do not allow you to flash custom firmware to them these days, especially for printers as there is money at stake (you could write firmware to support non-vendors Ink/Toner cartridges), or to enable disabled features.

In addition printers contain custom specific IP blocks related to everything from ethernet to scan and print functionality. It is just not feasible for someone to be able to write custom firmware for a printer without access to the hardware documentation for these devices. More sophisticated printers even have multiple CPUs for doing various tasks.

[adwn]

> Why would you design one from scratch? The hardware out there is excellent and inexpensive.

I thought the discussion was about open source printers, not about open source printer firmware? Sure, you could probably reverse engineer the interface to the actuators and sensors, which would allow you to write open source drivers, but you'd still be stuck with the printer manufacturer's cartridges.

[rmah]

The problem is not the design or the firmware. The problem is the production of the actual physical printer. Building real things is not like compiling software. Nothing happens exactly the same every time. The minimal exposure I had to manufacturing has shown me that even simple manufacturing steps are infuriatingly difficult to get consistent.